How can I increase precision of variables by the Coder toolbox generated c source code?
Unfortunately I have a ill-conditioned problem.
-
Did you mean ill-conditioned problem?– JohnCommented Jun 13, 2012 at 16:51
-
double precision is the standard for all sorts of scientific computing. If your problem is too ill-conditioned for a standard double-precision solver, then you need to fix your problem in the math world before you get to the computer world.– PeterCommented Jun 13, 2012 at 19:37
-
@Peter: maybe the OP has a numeric problem which is intrinsically hard and cannot resort to rounding because he looks for an exact solution. Happens in formal verification, unfortunately.– Luca GerettiCommented Jun 14, 2012 at 11:34
Add a comment
|
1 Answer
I think that unless you rely on an external library, you cannot work at higher precision than what long double allows. I'm not aware (but I could be wrong) of any ability to interface Coder to the Multiple Precision Toolbox types on native code.
I would suggest to start from the generated C source code and then use an arbitrary precision library such as apfloat. It requires some extra effort on your part, though.
answered Jun 13, 2012 at 17:24
